Radiographic features of knee and hip osteoarthritis represent characteristics of an individual, in addition to severity of osteoarthritis MB Kinds 1,2 , KL Vincken 2 , EP Vignon 3 , S ten Wolde 4 , JWJ Bijlsma 1 , PMJ Welsing 1,5 , ACA Marijnissen 1 , FPJG Lafeber 1 1 Rheumatology and Clinical Immunology, University Medical Centre (UMC) Utrecht, 2 Image Sciences Institute, UMC Utrecht, The Netherlands, 3 Rheumatology, Lyon-Sud University Hospital, Pierre-Bénite, France, 4 Rheumatology, Kennemer Gasthuis Haarlem, and 5 Julius Centre for Health Sciences and Primary Care, UMC Utrecht, The Netherlands Objective: To evaluate to what extent radiographic features of knees and hips that are normally related to osteoarthritis (OA) represent characteristics of an individual in addition to OA severity. Methods: We studied a cohort of individuals (n ¼ 1002) with very early signs of hip and knee OA, from the Cohort Hip and Cohort Knee (CHECK) study. Baseline radiographs were evaluated by digital analyses, using Holy’s and Knee Images Digital Analysis (KIDA) software, providing distinct quantitative measures of radiographic OA features. In addition, conventional Kellgren and Lawrence (KL) grading was performed. Digital parameters were evaluated for correlations within participants between contralateral (left vs. right hip and left vs. right knee), ipsilateral (e.g. left hip vs. left knee), and diagonal joints (e.g. left hip vs. right knee). Analyses were performed separately for participants with KL grade 0–I and those with evident radiographic OA (KL grade II–III). Regression analyses determined whether demographic characteristics were related to radiographic features. Results: Correlations between digital parameters and KL grade were moderate, and within each KL grade large variation was found. Within participants strong correlations were found for digital parameters between joints in individuals with KL grade 0–I (R ¼ 0.60–0.89), strongest for contralateral comparison, but no statistically significant correlations were found for participants with KL grade II–III. The demographic characteristics age, gender, height, and weight were, to a limited extent (R 2 ¼ 0.01–0.20) but statistically significant, related to radiographic characteristics. Conclusion: Using digital analyses of radiographic OA, strong correlations between joints within participants were found. These correlations diminished when OA became evident. This has implications for monitoring joint damage in (very) early OA with digital analyses. Osteoarthritis (OA) is a slowly developing joint disease characterized by pain and disability. Structural changes such as articular cartilage damage, osteophyte formation, synovial inflammation, and subchondral bone changes are assumed to give rise to these symptoms (1, 2). However, there is a discrepancy between radiographic and clinical features of OA, hampering definition, diagnosis, and eva- luation of progression (3). Despite this discrepancy and the progressive development of magnetic resonance imaging (MRI), which enables the detection of specific structural changes before becoming radiographically evident (4), radiographs are still considered the gold standard for demonstrating structural changes. This is because the image acquisition method is non-invasive, inexpensive, fast, and generally available (5, 6). Radiographic OA is commonly scored by use of a Kellgren and Lawrence (KL) grade (7). The drawback of such a grading is that it provides only a qualitative (ordinal) score of a combination of distinct structural aspects. In general, it takes up to 1 or 2 years before progression of a single radiographic grade becomes evi- dent (8, 9). Quantitative measurement of distinct radiographic features of OA by digital image analysis theoretically enables more precise measurement of disease severity with higher sensitivity to change. Therefore, measure- ment of joint space width (JSW) is frequently used as an individual quantitative parameter (10, 11). Quantitative measures of osteophyte area and bone density may also allow for the detection of small differences (12). The availability of these digital tools led us to evaluate to what extent radiographic features of knees and hips that are normally related to OA repre- sent characteristics of an individual, in addition to severity of OA. Floris PJ Lafeber, UMC Utrecht, Rheumatology and Clinical Immunology (F02.127), PO Box 85500, 3508 GA, Utrecht, The Netherlands. E-mail: f.lafeber@umcutrecht.nl Accepted 22 August 2011 Scand J Rheumatol 2012;41:141–149 141 © 2012 Taylor & Francis on license from Scandinavian Rheumatology Research Foundation DOI: 10.3109/03009742.2011.617311 www.scandjrheumatol.dk